xsurvey

Code for Individual Module "A Survey of Explainability Methods for Neural Network Classifiers in Natural Language Processing"

Getting Started

• create & activate environment and pip install -r requirements.txt

• install DeepExplain pip install -e git+https://github.com/marcoancona/DeepExplain.git#egg=deepexplain

• clone Repository

• copy sample.config.ini to config.ini

• set folders in config.ini

• create folders with python create_folders.py

• download dataset from https://www.kaggle.com/utathya/imdb-review-dataset and place it in the dataset folder specified

• download and unpack GloVe from http://nlp.stanford.edu/data/glove.840B.300d.zip and place it in the glove folder specified

Run code

• python preprocess.py

• python dist_plot.py

• python train_model.py

• python evaluation.py

• python plot.py -p -l, with -p being the option for creating plots and -l the option for creating colored tex files

• python correlation.py